WO2021004322A1 - Head special effect processing method and apparatus, and storage medium - Google Patents

Abstract

The present disclosure provides a head special effect processing method and apparatus, and a storage medium. The method comprises: obtaining an initial image; then recognizing a face image comprised in the initial image; accordingly obtaining an expanded image of an area where the face image is located from the initial image as a head area image; further processing the head area image by using a trained head recognition model to obtain a head image; then performing special effect processing on the head image to obtain a head special effect image; and displaying the head special effect image on the current display interface. According to the method of the present disclosure, the recognition accuracy for the head in a human image is improved, and the scenario in which a visual special effect is applied to the head is expanded.

Description

Translated fromChinese

头部特效处理方法及装置、存储介质Head special effect processing method, device and storage medium

相关申请的交叉引用Cross references to related applications

本申请要求于2019年07月09日提交的，申请号为201910615641.0、发明名称为“头部特效处理方法及装置、存储介质”的中国专利申请的优先权，该申请的全文通过引用结合在本申请中。This application claims the priority of the Chinese patent application filed on July 9, 2019 with the application number 201910615641.0 and the invention title "head special effect processing method and device, storage medium". The full text of this application is incorporated herein by reference. Applying.

技术领域Technical field

本公开涉及计算机技术领域，尤其涉及一种头部特效处理方法及装置、存储介质。The present disclosure relates to the field of computer technology, and in particular to a method and device for processing head special effects, and a storage medium.

背景技术Background technique

随着智能终端的普及与视觉特效技术的不断发展，越来越多的应用程序关注于为用户提供更加个性化与更佳视觉感受的视觉特效。With the popularization of smart terminals and the continuous development of visual effects technology, more and more applications focus on providing users with more personalized and better visual effects.

目前，针对人像头部的视觉特效一般关注于人脸，而非整个头部。现有技术中，通过人脸识别技术来识别出图像中的人脸，并对人脸进行特效处理，进而在终端屏幕上展示视觉特效。At present, visual effects for the head of a portrait generally focus on the face, rather than the entire head. In the prior art, a face recognition technology is used to recognize a face in an image, and special effect processing is performed on the face, and then visual special effects are displayed on the terminal screen.

由于人脸识别技术仅能实现对人脸的识别，而无法得到完整的头部图像，例如，无法包含人像的耳朵和头发等，如此，在无法识别出整个头部图像的前提下，也就无法对整个头部进行特效处理。Because the face recognition technology can only realize the recognition of the face, it cannot obtain the complete head image, for example, the ears and hair of the portrait cannot be included. Therefore, under the premise that the entire head image cannot be recognized, Cannot perform special effects on the entire head.

发明内容Summary of the invention

本公开提供一种头部特效处理方法及装置、存储介质，用以提高针对人像头部的识别准确度，并扩展视觉特效应用于头部的应用场景。The present disclosure provides a head special effect processing method, a device, and a storage medium, which are used to improve the recognition accuracy for the head of a portrait, and expand the visual special effect for application scenarios of the head.

第一方面，本公开提供一种头部特效处理方法，包括：In a first aspect, the present disclosure provides a head special effect processing method, including:

获取初始图像；Get the initial image;

识别所述初始图像中包含的人脸图像；Identifying the face image contained in the initial image;

在所述初始图像中，获取所述人脸图像所在区域的扩大图像，以作为头部区域图像；In the initial image, acquiring an enlarged image of the area where the face image is located as a head area image;

利用训练好的头部识别模型处理所述头部区域图像，得到头部图像；Use the trained head recognition model to process the head region image to obtain a head image;

对所述头部图像进行特效处理，得到头部特效图像；Performing special effect processing on the head image to obtain a head special effect image;

在当前显示界面上展示所述头部特效图像。The head special effect image is displayed on the current display interface.

第二方面，本公开提供一种头部特效处理装置，包括：In a second aspect, the present disclosure provides a head special effects processing device, including:

第一获取模块，用于获取初始图像；The first acquisition module is used to acquire the initial image;

第一识别模块，用于识别所述初始图像中包含的人脸图像；The first recognition module is configured to recognize the face image contained in the initial image;

第二获取模块，用于在所述初始图像中，获取所述人脸图像所在区域的扩大图像，以作为头部区域图像；A second acquisition module, configured to acquire an enlarged image of the region where the face image is located in the initial image, as a head region image;

第二识别模块，用于利用训练好的头部识别模型处理所述头部区域图像，得到头部图像；The second recognition module is used to process the image of the head region by using the trained head recognition model to obtain a head image;

特效处理模块，用于对所述头部图像进行特效处理，得到头部特效图像；A special effect processing module for performing special effect processing on the head image to obtain a head special effect image;

展示模块，用于在当前显示界面上展示所述头部特效图像。The display module is used to display the head special effect image on the current display interface.

第三方面，本公开提供一种头部特效处理装置，包括：In a third aspect, the present disclosure provides a head special effects processing device, including:

存储器；Memory

处理器；以及Processor; and

计算机程序；Computer program;

其中，所述计算机程序存储在所述存储器中，并被配置为由所述处理器执行以实现如第一方面所述的方法。Wherein, the computer program is stored in the memory and configured to be executed by the processor to implement the method according to the first aspect.

第五方面，本公开提供一种计算机可读存储介质，其上存储有计算机程序，In a fifth aspect, the present disclosure provides a computer-readable storage medium on which a computer program is stored,

所述计算机程序被处理器执行以实现如第一方面所述的方法。The computer program is executed by the processor to implement the method according to the first aspect.

本公开提供的一种头部特效处理方法及装置、存储介质，通过对初始图像进行人脸识别得到人脸图像，进而，将其所在区域扩大，即可得到包含人像完整头部和背景图像的头部区域图像，而考虑到其中还存在背景图像等非头部图像，因此，本公开还进一步利用训练好的头部识别模型对其进行处理，如此，即可得到精确的头部图像，识别准确度较高，解决了现有技术中仅能够识别人脸而无法识别头部的情况；从而，在此基础上对头部图像进行视觉特效处理并进行展示，视觉特效无需再局限于脸部，而是可以针对头部实现不同的视觉特效，扩展了视觉特效应用于头部这一具体场景。The present disclosure provides a head special effect processing method, device, and storage medium. The face image is obtained by performing face recognition on the initial image, and then, by expanding the area where it is located, a complete head and background image containing the portrait can be obtained. The head region image, and considering that there are also non-head images such as background images, the present disclosure further uses a trained head recognition model to process it. In this way, an accurate head image can be obtained. The accuracy is high, which solves the situation that the existing technology can only recognize the face but not the head; thus, on this basis, the head image is processed and displayed with visual special effects, and the visual special effects are no longer limited to the face , But can achieve different visual effects for the head, expanding the visual effects for the specific scene of the head.

附图说明Description of the drawings

此处的附图被并入说明书中并构成本说明书的一部分，示出了符合本公 开的实施例，并与说明书一起用于解释本公开的原理。The drawings herein are incorporated into the specification and constitute a part of the specification, show embodiments conforming to the disclosure, and together with the specification are used to explain the principle of the disclosure.

图1为现有技术中的视觉特效显示效果示意图；Fig. 1 is a schematic diagram of a visual special effect display effect in the prior art;

图2为本公开实施例所提供的一种终端的架构示意图；FIG. 2 is a schematic diagram of the architecture of a terminal provided by an embodiment of the disclosure;

图3为本公开实施例所提供的一种头部特效处理方法的流程示意图；3 is a schematic flowchart of a method for processing head special effects provided by an embodiment of the disclosure;

图4为本公开实施例所提供的另一种头部特效处理方法的流程示意图；4 is a schematic flowchart of another head special effect processing method provided by an embodiment of the disclosure;

图5为本公开实施例所提供的另一种头部特效处理方法的流程示意图；FIG. 5 is a schematic flowchart of another head special effect processing method provided by an embodiment of the disclosure;

图6为本公开实施例所提供的另一种头部特效处理方法的流程示意图；FIG. 6 is a schematic flowchart of another head special effect processing method provided by an embodiment of the disclosure;

图7为图6所示方法中的对齐效果示意图；FIG. 7 is a schematic diagram of the alignment effect in the method shown in FIG. 6;

图8为本公开实施例所提供的另一种头部特效处理方法的流程示意图；FIG. 8 is a schematic flowchart of another head special effect processing method provided by an embodiment of the disclosure;

图9为本公开实施例所提供的一种头部特效处理装置的功能方块图；FIG. 9 is a functional block diagram of a head special effects processing device provided by an embodiment of the disclosure;

图10为本公开实施例所提供的一种头部特效处理装置的实体结构示意图；10 is a schematic diagram of the physical structure of a head special effects processing device provided by an embodiment of the disclosure;

图11为本公开实施例所提供的一种头部特效处理装置的实体结构示意图。FIG. 11 is a schematic diagram of the physical structure of a head special effect processing apparatus provided by an embodiment of the disclosure.

通过上述附图，已示出本公开明确的实施例，后文中将有更详细的描述。这些附图和文字描述并不是为了通过任何方式限制本公开构思的范围，而是通过参考特定实施例为本领域技术人员说明本公开的概念。Through the above-mentioned drawings, the specific embodiments of the present disclosure have been shown, which will be described in more detail below. The drawings and text description are not intended to limit the scope of the concept of the present disclosure in any way, but to explain the concept of the present disclosure for those skilled in the art by referring to specific embodiments.

具体实施方式Detailed ways

这里将详细地对示例性实施例进行说明，其示例表示在附图中。下面的描述涉及附图时，除非另有表示，不同附图中的相同数字表示相同或相似的要素。以下示例性实施例中所描述的实施方式并不代表与本公开相一致的所有实施方式。相反，它们仅是与如所附权利要求书中所详述的、本公开的一些方面相一致的装置和方法的例子。Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements. The implementation manners described in the following exemplary embodiments do not represent all implementation manners consistent with the present disclosure. Rather, they are merely examples of devices and methods consistent with some aspects of the present disclosure as detailed in the appended claims.

本公开具体的应用场景为：针对图像中人像的头部识别场景。更进一步的，还可以具体应用于：对图像中的头部进行视觉特效展示的场景。The specific application scenario of the present disclosure is: a head recognition scenario for a portrait in an image. Furthermore, it can also be specifically applied to the scene where the head in the image is displayed with special visual effects.

请参考图1，现有的头部特效处理方法一般仅能够对人脸的高准确率识别，而这种识别方式无法识别头部除人脸之外的其他部分，如人像的头发和耳朵，也就是，无法实现针对头部的高精度识别。此时，针对人像头部的视觉特效实质为针对人脸的视觉特效，而非整个头部，这也限制了视觉特效的 应用场景。Please refer to Figure 1. Existing head special effects processing methods generally can only recognize human faces with high accuracy, and this recognition method cannot recognize other parts of the head other than the human face, such as the hair and ears of a portrait. That is, high-precision recognition of the head cannot be realized. At this time, the visual special effects for the head of the portrait are essentially the visual special effects for the face, not the entire head, which also limits the application scenarios of the visual effects.

本公开提供的技术方案，旨在解决现有技术的如上技术问题，并提出如下解决思路：首先，利用人脸识别技术识别出人脸图像，然后，将人脸图像所在区域进行扩大，即可得到初始图像中的头部区域图像，并考虑到其中可能包含背景图像，因此，进一步利用神经网络模型来对其进行处理，以得到精确的头部图像。如此，即可在完整的头部图像上展示视觉特效。The technical solution provided by the present disclosure aims to solve the above technical problems of the prior art, and proposes the following solution ideas: first, the face image is recognized by the face recognition technology, and then the area where the face image is located is expanded. The head region image in the initial image is obtained, and considering that it may contain a background image, the neural network model is further used to process it to obtain an accurate head image. In this way, visual effects can be displayed on the complete head image.

本公开提供的头部特效处理方法，可以适用于图2所示的终端中。如图2所示，终端200中包含：终端主体210与头部特效处理装置700，其中，头部特效处理装置700用于执行该头部特效处理方法。The head special effect processing method provided by the present disclosure can be applied to the terminal shown in FIG. 2. As shown in FIG. 2, the terminal 200 includes a terminal body 210 and a head special effect processing device 700, where the head special effect processing device 700 is used to execute the head special effect processing method.

本公开实施例对于终端主体所包含的组件无特别限制。在一种实际的实现场景中，可以包括以下一个或多个组件：处理组件，存储器，电源组件，多媒体组件，音频组件，输入/输出(I/O)接口，传感器组件，以及通信组件。The embodiments of the present disclosure have no particular limitation on the components included in the terminal body. In an actual implementation scenario, it may include one or more of the following components: processing components, memory, power supply components, multimedia components, audio components, input/output (I/O) interfaces, sensor components, and communication components.

而本公开实施例所涉及到的终端可以是无线终端也可以是有线终端。无线终端可以是指向用户提供语音和/或其他业务数据连通性的设备，具有无线连接功能的手持式设备、或连接到无线调制解调器的其他处理设备。无线终端可以经无线接入网(Radio Access Network，简称RAN)与一个或多个核心网设备进行通信，无线终端可以是移动终端，如移动电话(或称为“蜂窝”电话)和具有移动终端的计算机，例如，可以是便携式、袖珍式、手持式、计算机内置的或者车载的移动装置，它们与无线接入网交换语言和/或数据。再例如，无线终端还可以是个人通信业务(Personal Communication Service，简称PCS)电话、无绳电话、会话发起协议(Session Initiation Protocol，简称SIP)话机、无线本地环路(Wireless Local Loop，简称WLL)站、个人数字助理(Personal Digital Assistant，简称PDA)等设备。无线终端也可以称为系统、订户单元(Subscriber Unit)、订户站(Subscriber Station)，移动站(Mobile Station)、移动台(Mobile)、远程站(Remote Station)、远程终端(Remote Terminal)、接入终端(Access Terminal)、用户终端(User Terminal)、用户代理(User Agent)、用户设备(User Device or User Equipment)，在此不作限定。可选的，上述终端设备还可以是智能手表、平板电脑等设备。The terminals involved in the embodiments of the present disclosure may be wireless terminals or wired terminals. A wireless terminal may be a device that provides voice and/or other service data connectivity to a user, a handheld device with a wireless connection function, or other processing devices connected to a wireless modem. A wireless terminal can communicate with one or more core network devices via a radio access network (Radio Access Network, RAN). The wireless terminal can be a mobile terminal, such as a mobile phone (or "cellular" phone) and a mobile terminal The computer, for example, may be a portable, pocket-sized, handheld, built-in computer or vehicle-mounted mobile device, which exchanges language and/or data with the wireless access network. For another example, the wireless terminal may also be a Personal Communication Service (PCS) phone, a cordless phone, a Session Initiation Protocol (SIP) phone, and a wireless local loop (Wireless Local Loop, WLL) station. , Personal Digital Assistant (PDA) and other equipment. Wireless terminals can also be called systems, subscriber units (Subscriber Unit), subscriber stations (Subscriber Station), mobile stations (Mobile Station), mobile stations (Mobile), remote stations (Remote Station), remote terminals (Remote Terminal), The access terminal (Access Terminal), user terminal (User Terminal), user agent (User Agent), and user equipment (User Device or User Equipment) are not limited here. Optionally, the aforementioned terminal device may also be a smart watch, a tablet computer, or other devices.

下面以具体地实施例对本公开的技术方案以及本申请的技术方案如何解 决上述技术问题进行详细说明。下面这几个具体的实施例可以相互结合，对于相同或相似的概念或过程可能在某些实施例中不再赘述。下面将结合附图，对本公开的实施例进行描述。Hereinafter, specific embodiments are used to describe in detail the technical solutions of the present disclosure and how the technical solutions of the present application solve the above technical problems. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present disclosure will be described below with reference to the accompanying drawings.

实施例一Example one

本公开实施例提供了一种头部特效处理方法。请参考图3与图4，该方法包括如下步骤：The embodiment of the present disclosure provides a head special effect processing method. Please refer to Figure 3 and Figure 4, the method includes the following steps:

S302，获取初始图像。S302: Acquire an initial image.

本公开实施例所涉及到的初始图像，可以为一帧图像或至少两帧图像，其中，至少两帧图像可以包含：连续的至少两帧图像，和/或，不连续的至少两帧图像。也就是，初始图像可以是静态的摄像头采集到的单张图像，亦可以是视频或其他连续播放多媒体文件中的图像。The initial image involved in the embodiments of the present disclosure may be one frame of image or at least two frames of images, where the at least two frames of images may include: at least two consecutive frames of images, and/or at least two discontinuous frames of images. That is, the initial image can be a single image collected by a static camera, or it can be a video or another image in a continuously played multimedia file.

本公开实施例为了便于说明，仅给出针对一帧初始图像的处理方式，而针对获取到的多帧图像的情况，则每一帧图像均可按照本公开实施例所提供的后续方法进行处理，不再赘述。For ease of description, the embodiment of the present disclosure only provides a processing method for one frame of the initial image, and for the case of acquired multiple frames of images, each frame of image can be processed according to the subsequent method provided by the embodiment of the present disclosure ,No longer.

该步骤所获取到的包含初始图像，可以是图像采集装置当前实时采集到的图像；或者，也可以是已经预先采集完毕并存储好的图像。也就是，本公开实施例所提供方法既能适用于实时图像的处理，也能够适用于对历史图像的处理。The initial image acquired in this step may be an image currently acquired by the image acquisition device in real time; or, it may also be an image that has been acquired and stored in advance. That is, the method provided by the embodiment of the present disclosure can be applied to both real-time image processing and historical image processing.

在具体在执行该获取步骤时，若是针对实时图像进行处理，则直接获取或接收图像采集装置当前实时采集到的图像以作为初始图像即可；若是针对历史图像进行处理，则可以根据用户操作确定用户指示的存储位置所存储的图像，并将其作为初始图像即可。When performing this acquisition step, if the real-time image is processed, the image currently acquired by the image acquisition device in real time can be directly acquired or received as the initial image; if it is processed for the historical image, it can be determined according to user operations The image stored in the storage location indicated by the user can be used as the initial image.

S304，识别所述初始图像中包含的人脸图像。S304: Recognizing a face image included in the initial image.

如图4所示，该步骤用以识别出初始图像中的人脸图像。As shown in Figure 4, this step is used to identify the face image in the initial image.

在具体实现本方案时，只要可实现识别图像中的人脸的技术均可用于该步骤，以获取初始图像中的人脸图像。具体的，本公开实施例至少给出如下实现方式：可以利用人脸识别技术与关键点识别技术中的至少一种，来识别人脸图像。When the solution is specifically implemented, as long as the technology that can recognize the face in the image can be used in this step, to obtain the face image in the initial image. Specifically, the embodiments of the present disclosure provide at least the following implementation manners: at least one of a face recognition technology and a key point recognition technology can be used to recognize a face image.

一种实现方式中，可以利用人脸识别模型来处理初始图像，并获取人脸识别模型输出的人脸图像即可。In one implementation manner, a face recognition model can be used to process the initial image and obtain the face image output by the face recognition model.

另一种实现方式中，可以按照标准人脸，对初始图像中的人脸进行人脸关键点的提取，并基于提取出来的人脸关键点，获取到人脸图像。In another implementation manner, the face key points can be extracted from the face in the initial image according to the standard face, and the face image can be obtained based on the extracted face key points.

S306，在所述初始图像中，获取所述人脸图像所在区域的扩大图像，以作为头部区域图像。S306: In the initial image, obtain an enlarged image of the area where the face image is located, as a head area image.

该步骤用于获取头部区域图像，在具体实现时，可以将人脸图像所在区域扩大N倍，其中，N为大于1的正数，如此，可以在初始图像上得到一个扩大后的区域，然后，在初始图像上获取该扩大后的区域对应的扩大图像，即可得到头部区域图像。This step is used to obtain the image of the head region. In specific implementation, the region where the face image is located can be enlarged by N times, where N is a positive number greater than 1, so that an enlarged region can be obtained on the initial image. Then, the expanded image corresponding to the expanded area is obtained on the initial image, and then the head area image can be obtained.

其中，人脸图像所在区域可以为：人脸图像的实际不规则区域，或者，亦可以为：人脸图像所在的矩形或圆形等规则区域。在实际实现场景中，根据实际需要执行即可。Wherein, the area where the face image is located may be: the actual irregular area of the face image, or it may also be: the rectangular or circular area where the face image is located. In the actual implementation scenario, it can be executed according to actual needs.

在具体实现场景中，头部区域图像相对于人脸图像，仅在初始图像上扩大了图像范围，而N值的确定关系到头部图像是否完全处于头部区域图像之内，因此，在具体实现时，可将N值设置为大于1.2的数值。一种实现场景中，N可以取值为1.5，也就是，头部区域图像的尺寸为人脸图像尺寸的1.5倍。In the specific implementation scenario, the head area image only expands the image range on the initial image compared to the face image, and the determination of the N value is related to whether the head image is completely within the head area image. Therefore, in the specific In implementation, the value of N can be set to a value greater than 1.2. In an implementation scenario, N may take a value of 1.5, that is, the size of the head region image is 1.5 times the size of the face image.

S308，利用训练好的头部识别模型处理所述头部区域图像，得到头部图像。S308: Use the trained head recognition model to process the head region image to obtain a head image.

通过该步骤处理后，可以得到初始图像中的头部图像。在此基础上，可以对头部图像进行特效处理。After processing through this step, the head image in the initial image can be obtained. On this basis, special effects can be processed on the head image.

S310，对所述头部图像进行特效处理，得到头部特效图像。S310: Perform special effect processing on the head image to obtain a head special effect image.

S312，在当前显示界面上展示所述头部特效图像。S312: Display the head special effect image on the current display interface.

也就是，在终端屏幕中显示出进行特效处理后的头部特效图像。That is, the special effect image of the head after the special effect processing is displayed on the terminal screen.

通过前述方式，本发明实施例通过对初始图像进行人脸识别得到人脸图像，进而，将其所在区域扩大，即可得到包含人像完整头部和背景图像的头部区域图像，而考虑到其中还存在背景图像等非头部图像，因此，本公开还进一步利用训练好的头部识别模型对其进行处理，如此，即可得到精确的头部图像，识别准确度较高，解决了现有技术中仅能够识别人脸而无法识别头部的情况；从而，在此基础上对头部图像进行视觉特效处理并进行展示，视觉特效无需再局限于脸部，而是可以针对头部实现不同的视觉 特效，扩展了视觉特效应用于头部这一具体场景。In the foregoing manner, the embodiment of the present invention obtains a face image by performing face recognition on the initial image, and then expands the area where it is located to obtain a head region image containing the complete head of the portrait and the background image, and considering that There are also non-head images such as background images. Therefore, the present disclosure further uses the trained head recognition model to process them. In this way, accurate head images can be obtained with high recognition accuracy, which solves the existing problems. The technology can only recognize the face but not the head; therefore, based on this, the head image is processed and displayed with visual special effects. The visual special effects are no longer limited to the face, but can be different for the head. The visual effects of, expand the visual effects for the specific scene of the head.

以下，对如图3所示方法中的各步骤的实现方式进行具体说明。Hereinafter, the implementation of each step in the method shown in FIG. 3 will be described in detail.

具体的，经S308步骤处理后，得到的头部图像示意图可以参考图4。需要说明的是，图4所示示意图为了表征头部图像的范围，以略微与实际头部图像不同尺寸的虚线示出，仅用于示意，并非表明识别出的头部图像与实际头部图像不符。Specifically, after processing in step S308, the obtained schematic diagram of the head image may refer to FIG. 4. It should be noted that, in order to characterize the range of the head image, the schematic diagram shown in Figure 4 is shown with a dashed line with a slightly different size from the actual head image. It is only used for illustration and does not indicate that the recognized head image and the actual head image Does not match.

具体的，本公开实施例对所涉及到的头部识别网络模型的类型无特别限制，可根据实际需要设计自定义的神经网络模型来实现。在一种可能的设计中，头部识别模型可以包括但不限于：卷积神经网络(Convolutional Neural Networks，CNN)模型，其中，所述CNN模型包括：Shuffle Net-v2模型。Specifically, the embodiments of the present disclosure have no particular limitation on the type of the head recognition network model involved, and can be implemented by designing a custom neural network model according to actual needs. In a possible design, the head recognition model may include, but is not limited to: a convolutional neural network (Convolutional Neural Networks, CNN) model, where the CNN model includes: a Shuffle Net-v2 model.

而头部识别模型在用于执行该步骤之前，该方法还可以包括如下步骤：获取样本数据；利用样本数据对初始模型进行训练，得到训练好的头部识别模型。模型训练过程在此不做赘述。Before the head recognition model is used to perform this step, the method may further include the following steps: obtaining sample data; using the sample data to train the initial model to obtain a trained head recognition model. The model training process will not be repeated here.

如此，本公开实施例中，首先通过识别人脸，在扩大其区域得到头部区域，进而，利用头部识别模型来识别出其中的头部图像，具备较高的准确度，解决了现有技术中仅能够识别人脸而无法识别头部的情况。In this way, in the embodiments of the present disclosure, firstly, the head region is obtained by expanding the area by recognizing the human face, and then using the head recognition model to recognize the head image, which has high accuracy and solves the existing problems. The technology can only recognize the face but not the head.

具体而言，头部识别模型的输出数据可以有多种不同的设计方式，本公卡实施例至少给出如下两种实现方案：Specifically, the output data of the head recognition model can have a variety of different design methods. The embodiment of the public card provides at least the following two implementation solutions:

一种实现方案中，头部识别模型的输出数据为：图像中各像素点的像素值。此时，需要对像素值做进一步处理，才能够得到头部图像。In an implementation scheme, the output data of the head recognition model is: the pixel value of each pixel in the image. At this point, further processing of the pixel value is needed to get the head image.

请参考图5所示流程，此时，S308可以通过如下方式实现：Please refer to the process shown in Figure 5. At this time, S308 can be implemented in the following ways:

S3082-2，将所述头部区域图像作为所述头部识别模型的输入，获取所述头部识别模型的输出，得到所述头部区域图像中各像素点的像素值。S3082-2: Use the head region image as the input of the head recognition model, obtain the output of the head recognition model, and obtain the pixel value of each pixel in the head region image.

此时，头部识别模型的输入为头部区域图像，输出为头部区域图像中各像素点的像素值。其中，各像素点的像素值表征的意义可以根据实际需要预设。At this time, the input of the head recognition model is the head region image, and the output is the pixel value of each pixel in the head region image. Among them, the meaning of the pixel value representation of each pixel can be preset according to actual needs.

一种实现场景中，头部识别模型输出的像素点的像素值可以预设为用于指示像素点是否为头部图像像素点的标识符，如A和B，其中，若像素值为A则表征该像素点不是头部像素点；反之，若像素值为B则表征该像素点是头部像素点。In an implementation scenario, the pixel value of the pixel output by the head recognition model can be preset as an identifier for indicating whether the pixel is a pixel of the head image, such as A and B, where if the pixel value is A, then It means that the pixel is not a head pixel; on the contrary, if the pixel value is B, it means that the pixel is a head pixel.

另一种实现场景中，头部识别模型输出的像素点的像素值可以为A～B之间的连续值，如此，在后续确定头部像素点时，还需要进一步将各像素值与预设像素阈值比对。In another implementation scenario, the pixel value of the pixel point output by the head recognition model can be a continuous value between A and B. In this way, when determining the head pixel point later, you need to further compare each pixel value with the preset Pixel threshold comparison.

其中，A和B可以任意取值。一种实现场景中，A可以为0，B可以为255。Among them, A and B can take arbitrary values. In an implementation scenario, A can be 0 and B can be 255.

S3082-4，根据各像素点的像素值，在各像素点中确定头部像素点。S3082-4: Determine the head pixel in each pixel according to the pixel value of each pixel.

在具体实现本方案时，如前所述，基于头部识别模型所输出的像素点的表征意义不同，则可以采取不同的确定策略。In the specific implementation of this solution, as mentioned above, based on the different characterization meanings of the pixels output by the head recognition model, different determination strategies can be adopted.

具体而言，若像素值为用于指示像素点是否为头部图像像素点的标识符，则可以直接根据像素值的表征意义来确定头部像素点。Specifically, if the pixel value is an identifier used to indicate whether the pixel is a pixel of a head image, the pixel of the head can be directly determined according to the characterization meaning of the pixel value.

若像素值为A～B之间的连续值，则可以与预设的像素阈值进行比对，来确定头部像素点。If the pixel value is a continuous value between A and B, it can be compared with the preset pixel threshold to determine the head pixel.

其中，预设的像素阈值可以根据实际需要自定义预设，而像素阈值可以为单向限定的阈值，例如，像素值大于或者等于下限像素阈值的像素点(或者还可预设为上限阈值，则获取像素值小于或者等于像素阈值的像素点)即为头部像素点；或者，限速阈值也可以预设为双向限定的阈值，此时，像素值在这两个双向限定阈值的范围内的像素点即为头部像素点。Among them, the preset pixel threshold can be customized according to actual needs, and the pixel threshold can be a unidirectionally limited threshold, for example, pixels whose pixel value is greater than or equal to the lower pixel threshold (or can also be preset as the upper threshold, Then the pixel value obtained is less than or equal to the pixel threshold) is the head pixel; or, the speed limit threshold can also be preset as a two-way limited threshold. At this time, the pixel value is within the range of the two two-way limited thresholds The pixels of is the head pixels.

S3082-6，根据所述头部像素点，获取所述头部图像。S3082-6: Acquire the head image according to the pixels of the head.

可知，头部像素点实际为头部图像的像素点，也即，将头部像素点组合在一起即可构成头部图像。因此，只需要在头部区域图像中提取出前述确定的头部像素点，即可得到头部图像。It can be seen that the head pixels are actually the pixels of the head image, that is, the head pixels can be combined to form the head image. Therefore, it is only necessary to extract the aforementioned determined head pixel points from the head region image to obtain the head image.

此时，得到的头部图像不再包含背景图像，其具体表现为不规则的头部图像。At this time, the obtained head image no longer contains the background image, which is specifically expressed as an irregular head image.

而这种实现方式，在头部识别模型之后，还尽一步结合了后续的逻辑判断，具备较高的准确度。In this way of implementation, after the head recognition model, the follow-up logical judgment is combined with one step, which has a higher accuracy.

另一种实现方案中，头部识别模型的输出数据即为：所述头部图像。此时，如图5所示，S308可以通过如下方式实现：In another implementation solution, the output data of the head recognition model is: the head image. At this time, as shown in Figure 5, S308 can be implemented as follows:

S3084，将所述头部区域图像作为所述头部识别模型的输入，获取所述头部识别模型的输出，得到所述头部图像。S3084: Use the head region image as the input of the head recognition model, and obtain the output of the head recognition model to obtain the head image.

也就是，头部识别模型的输入与输出都是图像，在具体实现过程中，仅 利用头部识别模型即可实现头部图像识别，处理步骤较少，有利于提高处理效率。That is, the input and output of the head recognition model are both images. In the specific implementation process, only the head recognition model can be used to realize head image recognition, with fewer processing steps, which is beneficial to improve processing efficiency.

除此之外，本公开实施例中，为了进一步提高前述识别方法的精确度，还进一步提供如下方案，请参考图6，该方法中在执行S308之前，还包括如下步骤：In addition, in the embodiments of the present disclosure, in order to further improve the accuracy of the foregoing identification method, the following solution is further provided. Please refer to FIG. 6. The method further includes the following steps before performing S308:

S307，对所述头部区域图像进行对齐处理。S307: Perform alignment processing on the image of the head region.

从而，在执行后续的S308步骤时，可以利用对齐后的所述头部区域图像来识别其中的头部图像。而经这种处理方式处理之后，头部识别模型输出的头部图像亦是对齐后的。Therefore, when performing the subsequent step S308, the aligned head region image can be used to identify the head image therein. After processing in this way, the head image output by the head recognition model is also aligned.

而本公开实施例所涉及到的对齐处理，在具体实现过程中，可表现为将倾斜的头部图像进行转正的处理。请参考图7，在进行对齐处理之前，获取到的头部区域图像的显示方向与终端的显示方向一致，但是，其中具体的头部图像的显示方向与终端的显示方向成一定角度，此时，可以通过该对齐处理，对头部区域图像进行对齐，使得对齐后的头部图像的显示方向能够与终端的显示方向一致，而对齐后的头部区域图像的显示方向则与终端的显示方向成一定角度。需要说明的是，图7仅用于演示说明头部变化，并未对人像的身体部分进行调整。However, the alignment processing involved in the embodiment of the present disclosure can be expressed as a processing of correcting the tilted head image in a specific implementation process. Please refer to Figure 7. Before the alignment process, the display direction of the acquired head region image is consistent with the display direction of the terminal, but the display direction of the specific head image is at a certain angle to the display direction of the terminal. , The head area image can be aligned through the alignment process, so that the display direction of the aligned head image can be consistent with the display direction of the terminal, and the display direction of the aligned head area image is the same as the display direction of the terminal At an angle. It should be noted that Fig. 7 is only used to demonstrate head changes, and does not adjust the body part of the portrait.

对头部区域图像进行对齐处理，能够简化头部识别模型的处理难度，并且，在一定程度上避免头部图像未转正对识别结果的不利影响，有利于提高识别结果的准确性。Aligning the image of the head region can simplify the processing difficulty of the head recognition model and, to a certain extent, avoid the adverse effect of the uncorrected head image on the recognition result, which is beneficial to improve the accuracy of the recognition result.

本公开实施例还进一步给出了对头部区域图像进行对齐操作的实现方式：The embodiment of the present disclosure further provides an implementation manner of performing an alignment operation on the head region image:

利用预设的标准人脸图像，获取所述人脸图像对应的对齐矩阵；Using a preset standard face image to obtain an alignment matrix corresponding to the face image;

利用所述对齐矩阵，对所述头部区域图像进行对齐。Using the alignment matrix to align the head region image.

其中，预设的标准人脸图像作为转正对齐的标准图像，可以据此获取当前人脸图像与该标准人脸图像之间的差异程度，并由此来获取所述人脸图像对应的对齐矩阵。而由于人脸图像与头部区域图像中的显示方向是一致的，因此，可以利用该对齐矩阵，来实现头部区域图像的对齐。Wherein, the preset standard face image is used as the standard image for right alignment, and the degree of difference between the current face image and the standard face image can be obtained accordingly, and the alignment matrix corresponding to the face image can be obtained accordingly . Since the display directions in the face image and the head region image are consistent, the alignment matrix can be used to achieve the alignment of the head region image.

更具体的一种实现中，对齐矩阵可以为仿射变换矩阵(affine_mat)。In a more specific implementation, the alignment matrix may be an affine transformation matrix (affine_mat).

基于前述任一种实现方式，均可实现针对头部图像的精确识别。在此基础上，本公开实施例还进一步给出了前述头部图像应用于视觉特效场景中的 实现方案。Based on any of the foregoing implementation methods, accurate recognition of head images can be achieved. On this basis, the embodiment of the present disclosure further provides an implementation scheme for applying the aforementioned head image to a visual special effect scene.

具体的，请参考图8，S310步骤在实现时可以包括如下步骤：Specifically, please refer to FIG. 8. Step S310 may include the following steps when implemented:

S3102，接收特效触发指令。S3102: Receive a special effect trigger instruction.

在具体的实现场景中，特效触发指令可以由用户通过在终端上的操作发出。本公开实施例中，可以持续检测用户在终端上的操作信息，并将该操作信息与预设的特效触发指令进行对比，若二者相匹配，则确定接收到特效触发指令，并执行后续特效处理。In a specific implementation scenario, the special effect trigger instruction can be issued by the user through an operation on the terminal. In the embodiment of the present disclosure, the user's operation information on the terminal can be continuously detected, and the operation information can be compared with the preset special effect trigger instruction. If the two match, the special effect trigger instruction is determined to be received, and the subsequent special effect is executed deal with.

本公开实施例对于接收到特效触发指令之前，终端是否输出其他视觉特效、终端是否采集或输出图像无特别限制，也就是，对本方案的触发场景无限制。The embodiments of the present disclosure have no special restrictions on whether the terminal outputs other visual special effects and whether the terminal collects or outputs images before receiving the special effect trigger instruction, that is, there are no restrictions on the trigger scene of the solution.

S3104，根据接收到的特效触发指令，对所述头部图像进行特效处理，得到头部特效图像。S3104: Perform special effect processing on the head image according to the received special effect trigger instruction to obtain a head special effect image.

具体实现时，可以通过确定所述特效触发指令所指示的预设视觉特效，然后，利用所述预设视觉特效处理所述头部图像，得到头部特效图像。In specific implementation, the preset visual special effect indicated by the special effect trigger instruction may be determined, and then the head image is processed by the preset visual special effect to obtain the head special effect image.

本公开实施例对特效处理方式的形式以及数目均无特别限定，在具体的实现场景中，可以根据需要来自定义设置。具体的实现场景中，所述特效处理可以包括但不限于如下至少一种：尺寸变更处理、颜色变更处理与添加装饰物处理。The embodiments of the present disclosure have no particular limitation on the form and number of special effect processing methods. In a specific implementation scenario, the settings can be customized according to needs. In a specific implementation scenario, the special effect processing may include but is not limited to at least one of the following: size change processing, color change processing, and decoration addition processing.

其中，尺寸变更处理可以包括：头部图像的尺寸变大或变小。Among them, the size change processing may include: the size of the head image becomes larger or smaller.

头部图像的尺寸变大可具体表现为：将头部图像放大，以在初始图像上形成一种头部被放大的视觉效果。The increase in the size of the head image can be specifically expressed as: the head image is enlarged to form a visual effect of the enlarged head on the initial image.

而头部图像的尺寸变小可具体表现为：将头部图像缩小，以在初始图像上形成一种头部被缩小的视觉效果，在这种视觉效果中，还需要额外考虑对头部图形缩小前后的空白区域的填补问题。一种可能的设计中，可以结合添加装饰物处理的方式，来在空白区域填充装饰物，如在空白区域添补花环或光圈或进行模糊化处理等。另一种可能的设计中，若在历史图像中能够获取到初始图像所在环境的背景图像，还可以按照背景图像对空白区域进行填充。The reduction in the size of the head image can be specifically expressed as: shrinking the head image to form a visual effect of the reduced head on the initial image. In this visual effect, additional consideration must be given to the head image. Fill in the blank area before and after shrinking. In a possible design, the method of adding decorations can be combined to fill the blank areas with decorations, such as adding wreaths or apertures or blurring the blank areas. In another possible design, if the background image of the environment where the initial image is located can be obtained in the historical image, the blank area can also be filled according to the background image.

颜色变更处理则可以根据特效触发指令，为头部图像来设置用户自定义的颜色，从而，呈现出一种头部变色的视觉效果。这种颜色变更可以针对整个头部图像，亦可将针对头部图像中的部分图像分别设计，如针对其中的人 脸区域进行变色，和/或，针对头发区域(头部图像中去除人脸图像的部分)进行变色。The color change processing can set a user-defined color for the head image according to the special effect trigger instruction, thereby presenting a visual effect of head color change. This color change can be applied to the entire head image, or it can be designed separately for part of the head image, such as changing the color of the face area, and/or for the hair area (the head image is removed from the face The part of the image) changes color.

添加装饰物处理亦可以根据特效触发指令，为头部图像来设置用户自定义或指示的装饰物，从而，呈现出一种装置头部的效果。本公开实施例所涉及到的装饰物可以包括但不限于如下至少一种：眼镜、帽子、光圈、饰品、模糊化装饰。The processing of adding decorations can also set user-defined or instructed decorations for the head image according to the special effect trigger instruction, thereby presenting a device head effect. The decorations involved in the embodiments of the present disclosure may include but are not limited to at least one of the following: glasses, hats, apertures, ornaments, and blurred decorations.

在如图8所示的实现方式中，在执行S312所示的在当前显示界面上展示所述头部特效图像时，至少可通过如下两种方式进行展示：In the implementation shown in FIG. 8, when the head special effect image is displayed on the current display interface as shown in S312, it can be displayed in at least the following two ways:

一种可能的设计中，可以将在所述初始图像中所述目标图像所处的位置，对所述头部特效图像进行渲染，从而，可直接在当前显示界面上展示渲染后的所述初始图像。In a possible design, the head special effect image may be rendered at the position of the target image in the initial image, so that the rendered initial image can be directly displayed on the current display interface. image.

这种实现方式是直接对初始图像进行修改，也就是，按照预设视觉特效对应的头部特效图像，在初始图像上渲染，得到最终展示的图像。This implementation method is to directly modify the initial image, that is, to render on the initial image according to the head special effect image corresponding to the preset visual special effect to obtain the final displayed image.

另一种可能的设计中，在当前显示界面上叠加显示所述头部特效图像与所述初始图像，其中，所述头部特效图像叠加显示在所述初始图像之上。In another possible design, the head special effect image and the initial image are superimposed and displayed on the current display interface, wherein the head special effect image is superimposed and displayed on the initial image.

这种实现方式极大地降低了对初始图像的修改或影响，头部特效图像与初始图像分离并通过叠加显示的方式展示视觉特效，这使得对头部特效图像的处理自由度更高，处理效率也更高。This implementation method greatly reduces the modification or influence on the initial image. The head special effect image is separated from the initial image and the visual special effect is displayed through superimposed display, which makes the processing of the head special effect image more free and efficient. Also higher.

此外，在前述识别头部图像过程中，还存在一种识别出的头部图像是经过对齐后的图像的实现方式，针对这种实现方式，还需要对其进行逆对齐处理，以使得最终呈现的头部特效图像能够与初始图像的显示角度相匹配。In addition, in the aforementioned process of recognizing the head image, there is also an implementation method in which the recognized head image is an aligned image. For this implementation method, it needs to be reverse-aligned to make the final presentation The special effect image of the head can match the display angle of the original image.

在这种实现场景中，还可以包括如下步骤：获取所述头部图像对应的对齐矩阵的逆矩阵，然后，利用所述对齐矩阵的逆矩阵，对所述头部图像或所述头部特效图像的显示角度进行调整。In this implementation scenario, it may also include the following steps: acquiring the inverse matrix of the alignment matrix corresponding to the head image, and then using the inverse matrix of the alignment matrix to apply the head image or the head special effect The display angle of the image is adjusted.

需要说明的是，该逆对齐处理的步骤，可以在S804执行之前，对识别出的头部图像进行逆对齐，并进而对逆对齐后的头部图像进行特效处理，如此保证由此得到的头部特效图像与初始图像的显示方向一致，视觉特效的展示效果较好。It should be noted that the steps of the inverse alignment processing may perform inverse alignment on the recognized head image before the execution of S804, and then perform special effects processing on the inversely aligned head image, so as to ensure the resulting head image The display direction of the special effect images is the same as that of the initial image, and the display effect of the visual special effects is better.

或者，该逆对齐处理的步骤，亦可以在执行S804之后而S806之前执行。例如，若通过叠加显示的方式展示头部特效图像，则可以在叠加之前对头部 特效图像的显示角度进行调整，之后，再进行叠加显示。Alternatively, the steps of the reverse alignment processing can also be performed after S804 is performed but before S806. For example, if the head special effect image is displayed through superimposition display, the display angle of the head special effect image can be adjusted before superimposing, and then superimposed display is performed afterwards.

可以理解的是，上述实施例中的部分或全部步骤或操作仅是示例，本申请实施例还可以执行其它操作或者各种操作的变形。此外，各个步骤可以按照上述实施例呈现的不同的顺序来执行，并且有可能并非要执行上述实施例中的全部操作。It can be understood that part or all of the steps or operations in the above-mentioned embodiments are only examples, and the embodiments of the present application may also perform other operations or variations of various operations. In addition, each step may be executed in a different order presented in the foregoing embodiment, and it may not be necessary to perform all operations in the foregoing embodiment.

本申请中使用的用词仅用于描述实施例并且不用于限制权利要求。如在实施例以及权利要求的描述中使用的，除非上下文清楚地表明，否则单数形式的“一个”(a)、“一个”(an)和“所述”(the)旨在同样包括复数形式。类似地，如在本申请中所使用的术语“和/或”是指包含一个或一个以上相关联的列出的任何以及所有可能的组合。另外，当用于本申请中时，术语“包括”(comprise)及其变型“包括”(comprises)和/或包括(comprising)等指陈述的特征、整体、步骤、操作、元素，和/或组件的存在，但不排除一个或一个以上其它特征、整体、步骤、操作、元素、组件和/或这些的分组的存在或添加。The terms used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and claims, unless the context clearly indicates otherwise, the singular forms of "a" (a), "one" (an) and "the" (the) are intended to also include plural forms . Similarly, the term "and/or" as used in this application refers to any and all possible combinations of one or more of the associated lists. In addition, when used in this application, the term "comprise" (comprise) and its variants "comprises" and/or including (comprising) and the like refer to the stated features, wholes, steps, operations, elements, and/or The existence of components does not exclude the existence or addition of one or more other features, wholes, steps, operations, elements, components, and/or groups of these.

以上所述，仅为本申请的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉本技术领域的技术人员在本申请揭露的技术范围内，可轻易想到变化或替换，都应涵盖在本申请的保护范围之内。因此，本申请的保护范围应以所述权利要求的保护范围为准。The above are only specific implementations of this application, but the protection scope of this application is not limited to this. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application. Should be covered within the scope of protection of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

实施例二Example two

基于上述实施例一所提供的头部特效处理方法，本公开实施例进一步给出实现上述方法实施例中各步骤及方法的装置实施例。Based on the head special effect processing method provided in the above-mentioned embodiment 1, the embodiment of the present disclosure further provides an embodiment of a device that implements each step and method in the above-mentioned method embodiment.

本公开实施例提供了一种头部特效处理装置，请参考图9，该头部特效处理装置900，包括：The embodiment of the present disclosure provides a head special effect processing device. Please refer to FIG. 9. The head special effect processing device 900 includes:

第一获取模块91，用于获取初始图像；Thefirst acquisition module 91 is used to acquire an initial image;

第一识别模块92，用于识别所述初始图像中包含的人脸图像；Thefirst recognition module 92 is configured to recognize the face image contained in the initial image;

第二获取模块93，用于在所述初始图像中，获取所述人脸图像所在区域的扩大图像，以作为头部区域图像；The second acquiringmodule 93 is configured to acquire, in the initial image, an enlarged image of the area where the face image is located, as a head area image;

第二识别模块94，用于利用训练好的头部识别模型处理所述头部区域图像，得到头部图像；Thesecond recognition module 94 is configured to process the head region image by using the trained head recognition model to obtain a head image;

特效处理模块95，用于对所述头部图像进行特效处理，得到头部特效图 像；The specialeffect processing module 95 is configured to perform special effect processing on the head image to obtain a head special effect image;

展示模块96，用于在当前显示界面上展示所述头部特效图像。Thedisplay module 96 is configured to display the head special effect image on the current display interface.

一种可能的设计中，所述第二识别模块94，具体用于：In a possible design, thesecond identification module 94 is specifically used for:

将所述头部区域图像作为所述头部识别模型的输入，获取所述头部识别模型的输出，得到所述头部区域图像中各像素点的像素值；Using the head region image as the input of the head recognition model, obtaining the output of the head recognition model, and obtaining the pixel value of each pixel in the head region image;

根据各像素点的像素值，在各像素点中确定头部像素点；Determine the head pixel in each pixel according to the pixel value of each pixel;

根据所述头部像素点，获取所述头部图像。Acquire the head image according to the pixels of the head.

另一种可能的设计中，所述第二识别模块94，具体用于：In another possible design, thesecond identification module 94 is specifically used for:

将所述头部区域图像作为所述头部识别模型的输入，获取所述头部识别模型的输出，得到所述头部图像。The head region image is used as the input of the head recognition model, and the output of the head recognition model is obtained to obtain the head image.

本公开实施例中，所述头部识别模型包括：卷积神经网络CNN模型；In the embodiment of the present disclosure, the head recognition model includes: a convolutional neural network CNN model;

所述CNN模型包括：Shuffle Net-v2模型。The CNN model includes: Shuffle Net-v2 model.

此外，在另一种可能的设计中，所述头部特效处理装置900还可以包括：对齐处理模块(图9未示出)；In addition, in another possible design, the head special effects processing device 900 may further include: an alignment processing module (not shown in FIG. 9);

所述对齐处理模块，用于在所述利用训练好的头部识别模型处理所述头部区域图像，得到头部图像之前，对所述头部区域图像进行对齐处理。The alignment processing module is configured to perform alignment processing on the head region image before the head region image is processed by using the trained head recognition model to obtain the head image.

具体的，所述对齐处理模块，具体用于：Specifically, the alignment processing module is specifically used for:

利用预设的标准人脸图像，获取所述人脸图像对应的对齐矩阵；Using a preset standard face image to obtain an alignment matrix corresponding to the face image;

利用所述对齐矩阵，对所述头部区域图像进行对齐。Using the alignment matrix to align the head region image.

另一种可能的设计中，所述第一识别模块92，具体用于：In another possible design, thefirst identification module 92 is specifically used for:

利用训练好的人脸识别模型处理所述初始图像，得到所述人脸图像。Use the trained face recognition model to process the initial image to obtain the face image.

此外，当具体应用于视觉特效场景中时，所述特效处理模块95，具体用于：In addition, when specifically applied to a visual special effect scene, the specialeffect processing module 95 is specifically used for:

接收特效触发指令；Receive special effect trigger instructions;

根据接收到的特效触发指令，对所述头部图像进行特效处理，得到头部特效图像。According to the received special effect trigger instruction, special effect processing is performed on the head image to obtain the head special effect image.

其中，一种具体的实现方案中，所述特效处理模块95具体用于：Among them, in a specific implementation scheme, the specialeffect processing module 95 is specifically configured to:

确定所述特效触发指令所指示的预设视觉特效；Determine the preset visual special effect indicated by the special effect trigger instruction;

利用所述预设视觉特效处理所述头部图像，得到头部特效图像。The head image is processed by using the preset visual special effect to obtain a head special effect image.

本公开实施例中所涉及到的所述特效处理可以包括但不限于如下至少一 种：尺寸变更处理、颜色变更处理与添加装饰物处理。The special effect processing involved in the embodiments of the present disclosure may include but is not limited to at least one of the following: size change processing, color change processing, and decoration addition processing.

其中，一种具体的实现方案中，所述展示模块96具体用于：Among them, in a specific implementation solution, thedisplay module 96 is specifically configured to:

在所述初始图像中所述目标图像所处的位置，对所述头部特效图像进行渲染；Rendering the head special effect image at the position of the target image in the initial image;

在当前显示界面上展示渲染后的所述初始图像。The rendered initial image is displayed on the current display interface.

另一种具体的实现方案中，所述展示模块96具体用于：In another specific implementation solution, thedisplay module 96 is specifically configured to:

在当前显示界面上叠加显示所述头部特效图像与所述初始图像，其中，所述头部特效图像叠加显示在所述初始图像之上。The head special effect image and the initial image are superimposed and displayed on the current display interface, wherein the head special effect image is superimposed and displayed on the initial image.

另一种具体的实现方案中，若所述头部图像为对齐后的图像，所述特效处理模块95，还可进一步用于：In another specific implementation solution, if the head image is an aligned image, the specialeffect processing module 95 may be further used for:

获取所述头部图像对应的对齐矩阵的逆矩阵；Acquiring the inverse matrix of the alignment matrix corresponding to the head image;

利用所述对齐矩阵的逆矩阵，对所述头部图像或所述头部特效图像的显示角度进行调整。Using the inverse matrix of the alignment matrix to adjust the display angle of the head image or the head special effect image.

图9所示实施例的头部特效处理装置900可用于执行上述方法实施例的技术方案，其实现原理和技术效果可以进一步参考方法实施例中的相关描述，可选的，该头部特效处理装置900可以为终端。The head special effect processing apparatus 900 of the embodiment shown in FIG. 9 can be used to implement the technical solutions of the above method embodiments. For its implementation principles and technical effects, please refer to the relevant descriptions in the method embodiments. Optionally, the head special effects processing The device 900 may be a terminal.

应理解以上图9所示头部特效处理装置900的各个模块的划分仅仅是一种逻辑功能的划分，实际实现时可以全部或部分集成到一个物理实体上，也可以物理上分开。且这些模块可以全部以软件通过处理元件调用的形式实现；也可以全部以硬件的形式实现；还可以部分模块以软件通过处理元件调用的形式实现，部分模块通过硬件的形式实现。例如，第二识别模块94可以为单独设立的处理元件，也可以集成在头部特效处理装置900中，例如终端的某一个芯片中实现，此外，也可以以程序的形式存储于头部特效处理装置900的存储器中，由头部特效处理装置900的某一个处理元件调用并执行以上各个模块的功能。其它模块的实现与之类似。此外这些模块全部或部分可以集成在一起，也可以独立实现。这里所述的处理元件可以是一种集成电路，具有信号的处理能力。在实现过程中，上述方法的各步骤或以上各个模块可以通过处理器元件中的硬件的集成逻辑电路或者软件形式的指令完成。It should be understood that the division of the various modules of the head special effects processing device 900 shown in FIG. 9 is only a division of logical functions, and may be fully or partially integrated into a physical entity in actual implementation, or may be physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be implemented in the form of hardware; part of the modules can be implemented in the form of software called by the processing elements, and some of the modules can be implemented in the form of hardware. For example, thesecond identification module 94 may be a separately established processing element, or it may be integrated in the head special effects processing device 900, such as implemented in a certain chip of the terminal. In addition, it may also be stored in the head special effects processing in the form of a program In the memory of the device 900, a certain processing element of the head special effects processing device 900 calls and executes the functions of the above modules. The implementation of other modules is similar. In addition, all or part of these modules can be integrated together or implemented independently. The processing element described here may be an integrated circuit with signal processing capability. In the implementation process, each step of the above method or each of the above modules can be completed by hardware integrated logic circuits in the processor element or instructions in the form of software.

例如，以上这些模块可以是被配置成实施以上方法的一个或多个集成电路，例如：一个或多个特定集成电路(Application Specific Integrated Circuit， ASIC)，或，一个或多个微处理器(digital singnal processor，DSP)，或，一个或者多个现场可编程门阵列(Field Programmable Gate Array，FPGA)等。再如，当以上某个模块通过处理元件调度程序的形式实现时，该处理元件可以是通用处理器，例如中央处理器(Central Processing Unit，CPU)或其它可以调用程序的处理器。再如，这些模块可以集成在一起，以片上系统(system-on-a-chip，SOC)的形式实现。For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as: one or more application specific integrated circuits (ASIC), or one or more microprocessors (digital singnal processor, DSP), or, one or more Field Programmable Gate Array (Field Programmable Gate Array, FPGA), etc. For another example, when one of the above modules is implemented in the form of a processing element scheduler, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processors that can call programs. For another example, these modules can be integrated together and implemented in the form of a system-on-a-chip (SOC).

并且，本公开实施例提供了一种头部特效处理装置，请参考图10，该头部特效处理装置900，包括：In addition, an embodiment of the present disclosure provides a head special effect processing device. Please refer to FIG. 10. The head special effect processing device 900 includes:

存储器910；Memory 910;

处理器920；以及Processor 920; and

计算机程序；Computer program;

其中，计算机程序存储在存储器910中，并被配置为由处理器920执行以实现如上述实施例所述的方法。Wherein, the computer program is stored in thememory 910 and is configured to be executed by theprocessor 920 to implement the method described in the foregoing embodiment.

其中，头部特效处理装置900中处理器920的数目可以为一个或多个，处理器920也可以称为处理单元，可以实现一定的控制功能。所述处理器920可以是通用处理器或者专用处理器等。在一种可选地设计中，处理器920也可以存有指令，所述指令可以被所述处理器920运行，使得所述头部特效处理装置900执行上述方法实施例中描述的方法。Wherein, the number ofprocessors 920 in the head special effect processing device 900 may be one or more, and theprocessors 920 may also be referred to as processing units, which may implement certain control functions. Theprocessor 920 may be a general-purpose processor or a special-purpose processor. In an optional design, theprocessor 920 may also store instructions, and the instructions may be executed by theprocessor 920 so that the head special effect processing device 900 executes the method described in the foregoing method embodiment.

在又一种可能的设计中，头部特效处理装置900可以包括电路，所述电路可以实现前述方法实施例中发送或接收或者通信的功能。In another possible design, the head special effects processing device 900 may include a circuit, which may implement the sending or receiving or communication function in the foregoing method embodiment.

可选地，所述头部特效处理装置900中存储器910的数目可以为一个或多个，存储器910上存有指令或者中间数据，所述指令可在所述处理器920上被运行，使得所述头部特效处理装置900执行上述方法实施例中描述的方法。可选地，所述存储器910中还可以存储有其他相关数据。可选地处理器920中也可以存储指令和/或数据。所述处理器920和存储器910可以单独设置，也可以集成在一起。Optionally, the number ofmemories 910 in the head special effects processing device 900 may be one or more, and instructions or intermediate data are stored in thememory 910, and the instructions may be executed on theprocessor 920 so that all The head special effect processing device 900 executes the method described in the foregoing method embodiment. Optionally, other related data may also be stored in thememory 910. Optionally, instructions and/or data may also be stored in theprocessor 920. Theprocessor 920 and thememory 910 may be provided separately or integrated together.

此外，如图10所示，在该头部特效处理装置900中还设置有收发器930，其中，所述收发器930可以称为收发单元、收发机、收发电路、或者收发器等，用于与测试设备或其他终端设备进行数据传输或通信，在此不再赘述。In addition, as shown in FIG. 10, the head special effects processing device 900 is also provided with atransceiver 930, where thetransceiver 930 may be called a transceiver unit, a transceiver, a transceiver circuit, or a transceiver, etc. Data transmission or communication with test equipment or other terminal equipment is not repeated here.

如图10所示，存储器910、处理器920与收发器930通过总线连接并通 信。As shown in Fig. 10, thememory 910, theprocessor 920 and thetransceiver 930 are connected and communicated via a bus.

若该头部特效处理装置900用于实现对应于图3中的方法时，处理器920用于完成相应的确定或者控制操作，可选的，还可以在存储器910中存储相应的指令。各个部件的具体的处理方式可以参考前述实施例的相关描述。If the head special effects processing device 900 is used to implement the method corresponding to FIG. 3, theprocessor 920 is used to complete corresponding determination or control operations, and optionally, corresponding instructions may be stored in thememory 910. For the specific processing manner of each component, reference may be made to the related description of the foregoing embodiment.

此外，在另一个可能的设计中，请参考图11，该头部特效处理装置900中还可以进一步设置：图像采集装置940与显示装置950；In addition, in another possible design, please refer to FIG. 11, the head special effects processing device 900 may be further provided: animage acquisition device 940 and adisplay device 950;

其中，图像采集装置940，用于采集所述初始图像；Wherein, theimage acquisition device 940 is used to acquire the initial image;

显示装置950，用于显示图像。Thedisplay device 950 is used for displaying images.

具体而言，可用于显示初始图像，亦可用于显示头部特效图像。Specifically, it can be used to display the initial image, and can also be used to display the head special effect image.

其中，图像采集装置940包括任意可以采集到多媒体图像的装置，如摄像头；而显示装置950的表现形式可以包括但不限于：终端屏幕、投影显示装置、其他与终端相连接的便携式显示装置等。Theimage collection device 940 includes any device that can collect multimedia images, such as a camera; and thedisplay device 950 may include, but is not limited to: a terminal screen, a projection display device, and other portable display devices connected to the terminal.

此外，本公开实施例提供了一种可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行以实现如实施例一所述的方法。In addition, an embodiment of the present disclosure provides a readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement the method as described in the first embodiment.

以及，本公开实施例提供了一种终端，请参考图2，该终端包括：头部特效处理装置900与终端主体210。And, an embodiment of the present disclosure provides a terminal. Please refer to FIG. 2. The terminal includes a head special effect processing device 900 and a terminal body 210.

其中，终端主体210中一般还配置有图像采集装置(如摄像头)与显示装置(如显示屏)等。此时，如图9所示的头部特效处理装置900中的图像采集装置和/或显示装置可以复用终端已有的设备。Among them, the terminal main body 210 is generally configured with an image acquisition device (such as a camera) and a display device (such as a display screen). At this time, the image acquisition device and/or the display device in the head special effects processing device 900 shown in FIG. 9 can reuse the existing equipment of the terminal.

由于本实施例中的各模块能够执行实施例一所示的方法，本实施例未详细描述的部分，可参考对实施例一的相关说明。Since each module in this embodiment can execute the method shown in Embodiment 1, for parts that are not described in detail in this embodiment, reference may be made to the related description of Embodiment 1.

本领域技术人员在考虑说明书及实践这里公开的发明后，将容易想到本公开的其它实施方案。本公开旨在涵盖本公开的任何变型、用途或者适应性变化，这些变型、用途或者适应性变化遵循本公开的一般性原理并包括本公开未公开的本技术领域中的公知常识或惯用技术手段。说明书和实施例仅被视为示例性的，本公开的真正范围和精神由下面的权利要求书指出。Those skilled in the art will easily think of other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptive changes of the present disclosure. These variations, uses, or adaptive changes follow the general principles of the present disclosure and include common knowledge or conventional technical means in the technical field not disclosed in the present disclosure. . The description and the embodiments are only regarded as exemplary, and the true scope and spirit of the present disclosure are pointed out by the following claims.

应当理解的是，本公开并不局限于上面已经描述并在附图中示出的精确结构，并且可以在不脱离其范围进行各种修改和改变。本公开的范围仅由所附的权利要求书来限制。It should be understood that the present disclosure is not limited to the precise structure that has been described above and shown in the drawings, and various modifications and changes can be made without departing from its scope. The scope of the present disclosure is only limited by the appended claims.

Claims (17)

Translated fromChinese

一种头部特效处理方法，其特征在于，包括：A method for processing head special effects, characterized in that it comprises:获取初始图像；Get the initial image;识别所述初始图像中包含的人脸图像；Identifying the face image contained in the initial image;在所述初始图像中，获取所述人脸图像所在区域的扩大图像，以作为头部区域图像；In the initial image, acquiring an enlarged image of the area where the face image is located as a head area image;利用训练好的头部识别模型处理所述头部区域图像，得到头部图像；Use the trained head recognition model to process the head region image to obtain a head image;对所述头部图像进行特效处理，得到头部特效图像；Performing special effect processing on the head image to obtain a head special effect image;在当前显示界面上展示所述头部特效图像。The head special effect image is displayed on the current display interface.根据权利要求1所述的方法，其特征在于，所述利用训练好的头部识别模型处理所述头部区域图像，得到头部图像，包括：The method according to claim 1, wherein the processing the head region image by using the trained head recognition model to obtain the head image comprises:将所述头部区域图像作为所述头部识别模型的输入，获取所述头部识别模型的输出，得到所述头部区域图像中各像素点的像素值；Using the head region image as the input of the head recognition model, obtaining the output of the head recognition model, and obtaining the pixel value of each pixel in the head region image;根据各像素点的像素值，在各像素点中确定头部像素点；Determine the head pixel in each pixel according to the pixel value of each pixel;根据所述头部像素点，获取所述头部图像。Acquire the head image according to the pixels of the head.根据权利要求1所述的方法，其特征在于，所述利用训练好的头部识别模型处理所述头部区域图像，得到头部图像，包括：The method according to claim 1, wherein the processing the head region image by using the trained head recognition model to obtain the head image comprises:将所述头部区域图像作为所述头部识别模型的输入，获取所述头部识别模型的输出，得到所述头部图像。The head region image is used as the input of the head recognition model, and the output of the head recognition model is obtained to obtain the head image.根据权利要求1-3任一项所述的方法，其特征在于，所述头部识别模型包括：卷积神经网络CNN模型；The method according to any one of claims 1-3, wherein the head recognition model comprises: a convolutional neural network CNN model;所述CNN模型包括：Shuffle Net-v2模型。The CNN model includes: Shuffle Net-v2 model.根据权利要求1-3任一项所述的方法，其特征在于，所述利用训练好的头部识别模型处理所述头部区域图像，得到头部图像之前，所述方法还包括：The method according to any one of claims 1 to 3, characterized in that, before the head region image is processed by using the trained head recognition model to obtain the head image, the method further comprises:对所述头部区域图像进行对齐处理。Aligning the image of the head region.根据权利要求5所述的方法，其特征在于，所述对所述头部区域图像进行对齐处理，包括：The method according to claim 5, wherein the aligning the image of the head region comprises:利用预设的标准人脸图像，获取所述人脸图像对应的对齐矩阵；Using a preset standard face image to obtain an alignment matrix corresponding to the face image;利用所述对齐矩阵，对所述头部区域图像进行对齐。Using the alignment matrix to align the head region image.根据权利要求1-3任一项所述的方法，其特征在于，所述识别所述初始图像中包含的人脸图像，包括：The method according to any one of claims 1 to 3, wherein the recognizing the face image contained in the initial image comprises:利用训练好的人脸识别模型处理所述初始图像，得到所述人脸图像。Use the trained face recognition model to process the initial image to obtain the face image.根据权利要求1所述的方法，其特征在于，所述对所述头部图像进行特效处理，包括：The method according to claim 1, wherein the performing special effects processing on the head image comprises:接收特效触发指令；Receive special effect trigger instructions;根据所述特效触发指令，对所述头部图像进行特效处理，得到所述头部特效图像。According to the special effect trigger instruction, perform special effect processing on the head image to obtain the head special effect image.根据权利要求8所述的方法，其特征在于，所述根据所述特效触发指令，对所述头部图像进行特效处理，得到头部特效图像，包括：8. The method according to claim 8, wherein the performing special effect processing on the head image according to the special effect trigger instruction to obtain the head special effect image comprises:确定所述特效触发指令所指示的预设视觉特效；Determine the preset visual special effect indicated by the special effect trigger instruction;利用所述预设视觉特效处理所述头部图像，得到头部特效图像。The head image is processed by using the preset visual special effect to obtain a head special effect image.根据权利要求8或9所述的方法，其特征在于，所述特效处理包括如下至少一种：尺寸变更处理、颜色变更处理与添加装饰物处理。The method according to claim 8 or 9, wherein the special effect processing includes at least one of the following: size change processing, color change processing, and decoration addition processing.根据权利要求8或9所述的方法，其特征在于，在当前显示界面上展示所述头部特效图像，包括：The method according to claim 8 or 9, wherein displaying the head special effect image on the current display interface comprises:在所述初始图像中所述目标图像所处的位置，对所述头部特效图像进行渲染；Rendering the head special effect image at the position of the target image in the initial image;在当前显示界面上展示渲染后的所述初始图像。The rendered initial image is displayed on the current display interface.根据权利要求8或9所述的方法，其特征在于，在当前显示界面上展示所述头部特效图像，包括：The method according to claim 8 or 9, wherein displaying the head special effect image on the current display interface comprises:在当前显示界面上叠加显示所述头部特效图像与所述初始图像，其中，所述头部特效图像叠加显示在所述初始图像之上。The head special effect image and the initial image are superimposed and displayed on the current display interface, wherein the head special effect image is superimposed and displayed on the initial image.根据权利要求8或9所述的方法，其特征在于，若所述头部图像为对齐后的图像，所述方法还包括：The method according to claim 8 or 9, wherein if the head image is an aligned image, the method further comprises:获取所述头部图像对应的对齐矩阵的逆矩阵；Acquiring the inverse matrix of the alignment matrix corresponding to the head image;利用所述对齐矩阵的逆矩阵，对所述头部图像或所述头部特效图像的显示角度进行调整。Using the inverse matrix of the alignment matrix to adjust the display angle of the head image or the head special effect image.一种头部特效处理装置，其特征在于，包括：A head special effects processing device, characterized in that it comprises:第一获取模块，用于获取初始图像；The first acquisition module is used to acquire the initial image;第一识别模块，用于识别所述初始图像中包含的人脸图像；The first recognition module is configured to recognize the face image contained in the initial image;第二获取模块，用于在所述初始图像中，获取所述人脸图像所在区域的扩大图像，以作为头部区域图像；A second acquisition module, configured to acquire an enlarged image of the region where the face image is located in the initial image, as a head region image;第二识别模块，用于利用训练好的头部识别模型处理所述头部区域图像，得到头部图像；The second recognition module is used to process the image of the head region by using the trained head recognition model to obtain a head image;特效处理模块，用于对所述头部图像进行特效处理，得到头部特效图像；A special effect processing module for performing special effect processing on the head image to obtain a head special effect image;展示模块，用于在当前显示界面上展示所述头部特效图像。The display module is used to display the head special effect image on the current display interface.一种头部特效处理装置，其特征在于，包括：A head special effects processing device, characterized in that it comprises:存储器；Memory处理器；以及Processor; and计算机程序；Computer program;其中，所述计算机程序存储在所述存储器中，并被配置为由所述处理器执行以实现如权利要求1-13任一项所述的方法。Wherein, the computer program is stored in the memory and is configured to be executed by the processor to implement the method according to any one of claims 1-13.根据权利要求15所述的装置，其特征在于，所述装置还包括：The device according to claim 15, wherein the device further comprises:图像采集装置，用于采集所述初始图像；An image acquisition device for acquiring the initial image;显示装置，用于显示图像。The display device is used to display images.一种计算机可读存储介质，其特征在于，其上存储有计算机程序，A computer-readable storage medium, characterized in that a computer program is stored thereon,所述计算机程序被处理器执行以实现如权利要求1-13任一项所述的方法。The computer program is executed by a processor to implement the method according to any one of claims 1-13.

Images